In this demo, you will learn how Serverless Functions can be used to do facial recognition in the Telegram chat.

Here you can watch a recording of this demo:

• OpenShift 4.6 (or newer)
• Red Hat OpenShift Serverless - installed as an operator from OperatorHub (version >= 1.11.0)
  • To install on a cluster that does not have this available, clone the serverless operator repository, ensure that you are logged into your OpenShift cluster and run make images install
• Red Hat Integration - Camel K - installed as an operator from OperatorHub (version >= 1.2.0)

You will also need the following API keys and accounts

• Microsoft Faces API Key
• Telegram Bot API KEY
• An image registry account such as docker.io or quay.io

• Set the Telegram API Key in the following files

  • deploy/kamelet-broker.yaml
  • receiver/func.yaml
  • responder/func.yaml
  • responder/package.json

• Set the Microsoft Faces API Key in the following files

  • processor/func.yaml

oc apply -f deploy/broker.yaml

The demo uses Telegram to send images to a bot. A CamelSource is used to receive these images and convert them to a CloudEvent which can then be sent to a Knative event sink to be processed by functions. Install the Telegram Kamelet which handeles the creation of CamelSource and binding it to the Broker for you..

kubectl apply -f kamelet-broker.yaml

Export image registry that's being to used to store function images, if you haven't done so:

export FUNC_REGISTRY=docker.io/johndoe

Deploy receiver:

kn func deploy -p receiver

And connect it to the Broker:

oc apply -f deploy/trigger-receiver.yaml

Deploy processor:

kn func deploy -p processor

And connect it to the Broker:

oc apply -f deploy/trigger-processor.yaml

Deploy responder:

kn func deploy -p responder

And connect it to the Broker:

oc apply -f deploy/trigger-responder.yaml